BZPEERPapers, people, evidence and research paths
Menu

Discover

SearchFind papers, people, topics, institutions and opportunities from one place.GraphSee how papers, authors, topics and evidence connect.

Research tools

HomeContinue from the papers, people and topics that matter now.CollectionsTurn saved papers into reading lists, evidence maps and shareable context.ResearchReconstruct the path that led you from search to paper to question.CollaborationMove from a paper, topic or expert request into a focused thread.PublishCreate full-text research articles with metadata, formulas and versions.

Network

ExpertsFind specialists whose work and reviews explain why they can help.CommunitiesJoin research groups organized around papers, topics and shared evidence.

Opportunities

ListingsBrowse roles, calls and collaborations connected to real research context.

Account

Log inReturn to your saved papers, graph views and active threads.Create accountStart saving papers, building a research trail and joining teams.
Log inCreate account

Source author record

Hong Huang

Hong Huang appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record
math.DGmath.APArtificial IntelligenceMachine LearningNetworking and Internet ArchitectureComputer VisionCryptography and SecurityInformation RetrievalInformation Theorymath.ATmath.GTmath.ITphysics.acc-phSocial and Information NetworksSoftware Engineering

Catalog footprint

What is connected

22works
15topics
4close collaborators

Actions

Connect this record

Log in to claim
Open graphBrowse works

Research graph

See the researcher in context

Open full explorer

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

Published work

22 published item(s)

preprint2026arXiv

A generalised pre-training strategy for deep learning networks in semantic segmentation of remotely sensed images

In the segmentation of remotely sensed images, deep learning models are typically pre-trained using large image databases like ImageNet before fine-tuned on domain-specific datasets. However, the performance of these fine-tuned models is often hindered by the large domain gaps (i.e., differences in scenes and modalities) between ImageNet's images and remotely sensed images being processed. Therefore, many researchers have undertaken efforts to establish large-scale domain-specific image datasets for pre-training, aiming to enhance model performance. However, establishing such datasets is often challenging, requiring significant effort, and these datasets often exhibit limited generaliza-bility to other application scenarios. To address these issues, this study introduces a novel yet simple pre-training strategy designed to guide a model away from learning domain-specific features in a pre-training dataset during pre-training, thereby improving the generalisation ability of the pre-trained model. To evaluate the strategy's effectiveness, deep learning models are pre-trained on ImageNet and subsequently fine-tuned on four semantic segmentation datasets with diverse scenes and modalities, including iSAID, MFNet, PST900 and Potsdam. Experimental results show that the proposed pre-training strategy led to state-of-the-art accuracies on all four datasets, namely 67.4% mIoU for iSAID, 56.9% mIoU for MFNet, 84.22% mIoU for PST900, 91.88% mF1 for Potsdam. This research lays the groundwork for developing a unified foundation model applicable to both computer vision and remote sensing applications.

0 citations0 reviewsNo code linkedOpen access
preprint2026arXiv

Scalable Heterogeneous Graph Learning via Heterogeneous-aware Orthogonal Prototype Experts

Heterogeneous Graph Neural Networks(HGNNs) have advanced mainly through better encoders, yet their decoding/projection stage still relies on a single shared linear head, assuming it can map rich node embeddings to labels. We call this the Linear Projection Bottleneck: in heterogeneous graphs, contextual diversity and long-tail shifts make a global head miss fine semantics, overfit hub nodes, and underserve tail nodes. While Mixture-of-Experts(MoE) could help, naively applying it clashes with structural imbalance and risks expert collapse. We propose a Heterogeneous-aware Orthogonal Prototype Experts framework named HOPE, a plug-and-play replacement for the standard prediction head. HOPE uses learnable prototype-based routing to assign instances to experts by similarity, letting expert usage follow the natural long-tail distribution, and adds expert orthogonalization to encourage diversity and prevent collapse. Experiments on four real datasets show consistent gains across SOTA HGNN backbones with minimal overhead.

0 citations0 reviewsNo code linkedOpen access
preprint2026arXiv

Sherry: Hardware-Efficient 1.25-Bit Ternary Quantization via Fine-grained Sparsification

The deployment of Large Language Models (LLMs) on resource-constrained edge devices is increasingly hindered by prohibitive memory and computational requirements. While ternary quantization offers a compelling solution by reducing weights to {-1, 0, +1}, current implementations suffer from a fundamental misalignment with commodity hardware. Most existing methods must choose between 2-bit aligned packing, which incurs significant bit wastage, or 1.67-bit irregular packing, which degrades inference speed. To resolve this tension, we propose Sherry, a hardware-efficient ternary quantization framework. Sherry introduces a 3:4 fine-grained sparsity that achieves a regularized 1.25-bit width by packing blocks of four weights into five bits, restoring power-of-two alignment. Furthermore, we identify weight trapping issue in sparse ternary training, which leads to representational collapse. To address this, Sherry introduces Arenas, an annealing residual synapse mechanism that maintains representational diversity during training. Empirical evaluations on LLaMA-3.2 across five benchmarks demonstrate that Sherry matches state-of-the-art ternary performance while significantly reducing model size. Notably, on an Intel i7-14700HX CPU, our 1B model achieves zero accuracy loss compared to SOTA baselines while providing 25% bit savings and 10% speed up. The code is available at https://github.com/Tencent/AngelSlim .

0 citations0 reviewsNo code linkedOpen access
preprint2021arXiv

Multi-Task Representation Learning with Multi-View Graph Convolutional Networks

Link prediction and node classification are two important downstream tasks of network representation learning. Existing methods have achieved acceptable results but they perform these two tasks separately, which requires a lot of duplication of work and ignores the correlations between tasks. Besides, conventional models suffer from the identical treatment of information of multiple views, thus they fail to learn robust representation for downstream tasks. To this end, we tackle link prediction and node classification problems simultaneously via multi-task multi-view learning in this paper. We first explain the feasibility and advantages of multi-task multi-view learning for these two tasks. Then we propose a novel model named as MT-MVGCN to perform link prediction and node classification tasks simultaneously. More specifically, we design a multi-view graph convolutional network to extract abundant information of multiple views in a network, which is shared by different tasks. We further apply two attention mechanisms: view attention mechanism and task attention mechanism to make views and tasks adjust the view fusion process. Moreover, view reconstruction can be introduced as an auxiliary task to boost the performance of the proposed model. Experiments on real-world network datasets demonstrate that our model is efficient yet effective, and outperforms advanced baselines in these two tasks.

0 citations0 reviewsNo code linkedOpen access
preprint2021arXiv

Outlier-Resilient Web Service QoS Prediction

The proliferation of Web services makes it difficult for users to select the most appropriate one among numerous functionally identical or similar service candidates. Quality-of-Service (QoS) describes the non-functional characteristics of Web services, and it has become the key differentiator for service selection. However, users cannot invoke all Web services to obtain the corresponding QoS values due to high time cost and huge resource overhead. Thus, it is essential to predict unknown QoS values. Although various QoS prediction methods have been proposed, few of them have taken outliers into consideration, which may dramatically degrade the prediction performance. To overcome this limitation, we propose an outlier-resilient QoS prediction method in this paper. Our method utilizes Cauchy loss to measure the discrepancy between the observed QoS values and the predicted ones. Owing to the robustness of Cauchy loss, our method is resilient to outliers. We further extend our method to provide time-aware QoS prediction results by taking the temporal information into consideration. Finally, we conduct extensive experiments on both static and dynamic datasets. The results demonstrate that our method is able to achieve better performance than state-of-the-art baseline methods.

0 citations0 reviewsNo code linkedOpen access
preprint2019arXiv

Kähler-Ricci flow on homogeneous toric bundles

Assume that $X$ is a homogeneous toric bundle of the form $G^{\mathbb{C}}\times_{P,τ} F$ and is Fano, where $G$ is a compact semisimple Lie group with complexification $G^\mathbb{C}$, $P$ a parabolic subgroup of $G^\mathbb{C}$, $τ:P\rightarrow (T^m)^\mathbb{C}$ is a surjective homomorphism from $P$ to the algebraic torus $(T^m)^\mathbb{C}$, and $F$ is a compact toric manifold of complex dimension $m$. In this note we show that the normalized Kähler-Ricci flow on $X$ with a $G\times T^m$-invariant initial Kähler form in $c_1(X)$ converges, modulo the algebraic torus action, to a Kähler-Ricci soliton. This extends a previous work of X. H. Zhu. As a consequence we recover a result of Podestà-Spiro.

0 citations0 reviewsNo code linkedOpen access
preprint2016arXiv

Exotic $\mathbb{R}^4$'s and positive isotropic curvature

We show that no exotic $\mathbb{R}^4$ admits a complete Riemannian metric with uniformly positive isotropic curvature and with bounded geometry. This is essentially a corollary of the main result in [Hu1], and was stated in [Hu2] without proof. In the process of the proof we also show that the diffeomorphism type of an infinite connected sum of some connected smooth $n$-manifolds ($n\geq 2$) according to a locally finite graph does not depend on the gluing maps used.

0 citations0 reviewsNo code linkedOpen access
preprint2016arXiv

Four-orbifolds with positive isotropic curvature

We prove the following result: Let $(X,g_0)$ be a complete, connected 4-manifold with uniformly positive isotropic curvature and with bounded geometry. Then there is a finite collection $\mathcal{F}$ of manifolds of the form $\mathbb{S}^3 \times \mathbb{R} /G$, where $G$ is a discrete subgroup of the isometry group of the round cylinder $\mathbb{S}^3\times \mathbb{R}$ on which $G$ acts freely, such that $X$ is diffeomorphic to a possibly infinite connected sum of $\mathbb{S}^4,\mathbb{RP}^4$ and members of $\mathcal{F}$. This extends recent work of Chen-Tang-Zhu and Huang. We also extend the above result to the case of orbifolds. The proof uses Ricci flow with surgery on complete orbifolds.

0 citations0 reviewsNo code linkedOpen access
preprint2016arXiv

Notes on branched coverings of Seifert manifolds

In a paper published in 2002, the author gave a criterion to determine whether there is a fiber-preserving branched covering between two given orientable Seifert manifolds with orientable bases. Here we supply some details of the proof of two claims in that paper. We give an explicit construction of fiber-preserving branched covering between two Seifert fibered solid tori when their Seifert invariants satisfy certain relation, and we show the factorability of fiber-preserving branched coverings between two closed Seifert manifolds.

0 citations0 reviewsNo code linkedOpen access
preprint2015arXiv

The Cauchy problem for fully nonlinear parabolic systems on manifolds

We show the short time existence and uniqueness of solutions to the Cauchy problem for fully nonlinear systems of arbitrary even order on closed manifolds which are strongly parabolic at the initial values. The proof uses a linearization procedure and a fixed-point argument, and the key ingredient is the well known Schauder estimates for linear, strongly parabolic systems.

0 citations0 reviewsNo code linkedOpen access
preprint2015arXiv

The transverse Chern-Ricci flow

We introduce transverse Chern-Ricci flow for transversely Hermitian foliations, which is analogous to the Chern-Ricci flow. We show that when $\mathcal{F}$ is homologically orientable and the basic first Bott-Chern class is zero, starting at any transversely Hermitian metric the flow exists for all time and as $t\rightarrow \infty$ converges smoothly to a transversely Hermitian metric $ω_\infty$ with the transverse Chern-Ricci form $ρ^T(ω_\infty)=0$. We also determine the maximal existence time of the flow in the general case. These are foliated version of results of Gill and Tosatti-Weinkove, and also extend recent work of Bedulli-He-Vezzoni.

0 citations0 reviewsNo code linkedOpen access
preprint2014arXiv

Applications of Compressed Sensing in Communications Networks

This paper presents a tutorial for CS applications in communications networks. The Shannon's sampling theorem states that to recover a signal, the sampling rate must be as least the Nyquist rate. Compressed sensing (CS) is based on the surprising fact that to recover a signal that is sparse in certain representations, one can sample at the rate far below the Nyquist rate. Since its inception in 2006, CS attracted much interest in the research community and found wide-ranging applications from astronomy, biology, communications, image and video processing, medicine, to radar. CS also found successful applications in communications networks. CS was applied in the detection and estimation of wireless signals, source coding, multi-access channels, data collection in sensor networks, and network monitoring, etc. In many cases, CS was shown to bring performance gains on the order of 10X. We believe this is just the beginning of CS applications in communications networks, and the future will see even more fruitful applications of CS in our field.

0 citations0 reviewsNo code linkedOpen access
preprint2014arXiv

Complete 4-manifolds with uniformly positive isotropic curvature

We prove the following result: Let $(X,g_0)$ be a complete, connected 4-manifold with uniformly positive isotropic curvature and with bounded geometry. Then there is a finite collection $\mathcal{F}$ of manifolds of the form $\mathbb{S}^3 \times \mathbb{R} /G$, where $G$ is a fixed point free discrete subgroup of the isometry group of the standard metric on $\mathbb{S}^3\times \mathbb{R}$, such that $X$ is diffeomorphic to a (possibly infinite) connected sum of copies of $\mathbb{S}^4,\mathbb{RP}^4$ and/or members of $\mathcal{F}$. This extends recent work of Chen-Tang-Zhu and Huang. We also extend the above result to the case of orbifolds. The proof uses Ricci flow with surgery on complete orbifolds.

0 citations0 reviewsNo code linkedOpen access
preprint2014arXiv

Network Multiple-Input and Multiple-Output for Wireless Local Area Networks

This paper presents a tutorial for network multiple-input and multiple-output (netMIMO) in wireless local area networks (WLAN). Wireless traffic demand is growing exponentially. NetMIMO allows access points (APs) in a WLAN to cooperate in their transmissions as if the APs form a single virtual MIMO node. NetMIMO can significantly increase network capacity by reducing interferences and contentions through the cooperation of the APs. This paper covers a few representative netMIMO methods, ranging from interference alignment and cancelation, channel access protocol to allow MIMO nodes to join ongoing transmissions, distributed synchronization, to interference and contention mitigation in multiple contention domains. We believe the netMIMO methods described here are just the beginning of the new technologies to address the challenge of ever-increasing wireless traffic demand, and the future will see even more new developments in this field.

0 citations0 reviewsNo code linkedOpen access
preprint2014arXiv

Network Traffic Anomaly Detection

This paper presents a tutorial for network anomaly detection, focusing on non-signature-based approaches. Network traffic anomalies are unusual and significant changes in the traffic of a network. Networks play an important role in today's social and economic infrastructures. The security of the network becomes crucial, and network traffic anomaly detection constitutes an important part of network security. In this paper, we present three major approaches to non-signature-based network detection: PCA-based, sketch-based, and signal-analysis-based. In addition, we introduce a framework that subsumes the three approaches and a scheme for network anomaly extraction. We believe network anomaly detection will become more important in the future because of the increasing importance of network security.

0 citations0 reviewsNo code linkedOpen access
preprint2013arXiv

A matrix differential Harnack estimate for a class of ultraparabolic equations

Let $u$ be a positive solution of the ultraparabolic equation \begin{equation*} \partial_t u=\sum_{i=1}^n \partial_{x_i}^2 u+\sum_{i=1}^k x_i\partial_{x_{n+i}}u \hspace{8mm} \mbox{on} \hspace{4mm} \mathbb{R}^{n+k}\times (0,T), \end{equation*} where $1\leq k\leq n$ and $0<T \leq +\infty$. Assume that $u$ and its derivatives (w.r.t. the space variables) up to the second order are bounded on any compact subinterval of $(0,T)$. Then the difference $H(\log u)- H(\log f)$ of the Hessian matrices of $\log u$ and of $\log f$ (both w.r.t. the space variables) is non-negatively definite, where $f$ is the fundamental solution of the above equation with pole at the origin $(0,0)$. The estimate in the case $n=k=1$ is due to Hamilton. As a corollary we get that $Δl+\frac{n+3k}{2t}+\frac{6k}{t^3}\geq 0$, where $l=\log u$, and $Δ=\sum_{i=1}^{n+k} \partial_{x_i}^2 $.

0 citations0 reviewsNo code linkedOpen access
preprint2013arXiv

Design of a 325MHz Beta=0.12 superconducting single spoke cavity for China-ADS

Twelve superconducting single spoke cavities whose Beta is 0.12 (Spoke012) operating at 325MHz, are adopted in Injector I for China-ADS linac. This type of spoke cavity is believed to be one of the key challenges for its very low geometry Beta. So far, the prototype cavity has been designed, fabricated and tested successfully. The design work was finished by using CST-MWS and ANSYS software for the RF and mechanical properties optimization. This paper presents the details of the final design for Spoke012 prototype cavity.

0 citations0 reviewsNo code linkedOpen access
preprint2013arXiv

Sasaki manifolds with positive transverse orthogonal bisectional curvature

In this short note we show the following result: Let $(M^{2n+1},g)$ ($n \geq 2$) be a compact Sasaki manifold with positive transverse orthogonal bisectional curvature. Then $π_1(M)$ is finite, and the universal cover of $(M^{2n+1},g)$ is isomorphic to a weighted Sasaki sphere. We also get some results in the case of nonnegative transverse orthogonal bisectional curvature under some additional conditions. This extends recent work of He and Sun. The proof uses Sasaki-Ricci flow.

0 citations0 reviewsNo code linkedOpen access
preprint2012arXiv

Three-orbifolds with positive scalar curvature

We prove the following result: Let $(\mathcal{O},g_0)$ be a complete, connected 3-orbifold with uniformly positive scalar curvature, with bounded geometry, and containing no bad 2-suborbifolds. Then there is a finite collection $\mathcal{F}$ of spherical 3-orbifolds, such that $\mathcal{O}$ is diffeomorphic to a (possibly infinite) orbifold connected sum of copies of members in $\mathcal{F}$. This extends work of Perelman and Bessi$\grave{e}$res-Besson-Maillot. The proof uses Ricci flow with surgery on complete 3-orbifolds, and are along the lines of the author's previous work on 4-orbifolds with positive isotropic curvature.

0 citations0 reviewsNo code linkedOpen access
preprint2011arXiv

Normal curvature bounds along the mean curvature flow

Let $(M^n,g_0)$ and $(\bar{M}^{n+1},\bar{g})$ be complete Riemannian manifolds with $|\bar{\nabla}^k\bar{Rm}|\le \bar{C}$ for $k \le 2$, and suppose there is an isometric immersion $F_0: M^n \rightarrow \bar{M}^{n+1}$ with bounded second fundamental form. Let $F_t: M^n \rightarrow \bar{M}^{n+1}$ ($t\in [0,T]$) be a family of immersions evolving by mean curvature flow with initial data $F_0$ and with uniformly bounded second fundamental forms. We show that the supremum and infimum of the normal curvature of the immersions $F_t$ vary at a bounded rate. This is an analogue of a result of Rong and Kapovitch on Ricci flow.

0 citations0 reviewsNo code linkedOpen access
preprint2011arXiv

Ricci flow on open 4-manifolds with positive isotropic curvature

In this note we prove the following result: Let $X$ be a complete, connected 4-manifold with uniformly positive isotropic curvature, with bounded geometry and with no essential incompressible space form. Then $X$ is diffeomorphic to $\mathbb{S}^4$, or $\mathbb{RP}^4$, or $\mathbb{S}^3\times \mathbb{S}^1$, or $\mathbb{S}^3\widetilde{\times} \mathbb{S}^1$, or a possibly infinite connected sum of them. This extends work of Hamilton and Chen-Zhu to the noncompact case. The proof uses Ricci flow with surgery on complete 4-manifolds, and is inspired by recent work of Bessi$\grave{e}$res, Besson and Maillot.

0 citations0 reviewsNo code linkedOpen access